Targeting transcription-coupled nucleotide excision repair ... - Nature

Leukemia (2017) 31, 1177–1186 © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved 0887-6924/17 www.nature.com/leu

ORIGINAL ARTICLE

Targeting transcription-coupled nucleotide excision repair overcomes resistance in chronic lymphocytic leukemia G Lohmann1,2, E Vasyutina1,2, J Bloehdorn3, N Reinart1,2, JI Schneider2,4, V Babu2,4, G Knittel1,2, G Crispatzu1,2, P Mayer1,2, C Prinz1,2, JK Muenzner5, B Biersack5, DG Efremov6, L Chessa7, CD Herling1,2, S Stilgenbauer3, M Hallek1,2, R Schobert5, HC Reinhardt1,2, B Schumacher2,4 and M Herling1,2 Treatment resistance becomes a challenge at some point in the course of most patients with chronic lymphocytic leukemia (CLL). This applies to fludarabine-based regimens, and is also an increasing concern in the era of more targeted therapies. As cells with low-replicative activity rely on repair that triggers checkpoint-independent noncanonical pathways, we reasoned that targeting the nucleotide excision repair (NER) reaction addresses a vulnerability of CLL and might even synergize with fludarabine, which blocks the NER gap-filling step. We interrogated here especially the replication-independent transcription-coupled-NER ((TC)-NER) in prospective trial patients, primary CLL cultures, cell lines and mice. We screen selected (TC)-NER-targeting compounds as experimental (illudins) or clinically approved (trabectedin) drugs. They inflict transcription-stalling DNA lesions requiring TC-NER either for their removal (illudins) or for generation of lethal strand breaks (trabectedin). Genetically defined systems of NER deficiency confirmed their specificity. They selectively and efficiently induced cell death in CLL, irrespective of high-risk cytogenetics, IGHV status or clinical treatment history, including resistance. The substances induced ATM/p53-independent apoptosis and showed marked synergisms with fludarabine. Trabectedin additionally perturbed stromal-cell protection and showed encouraging antileukemic profiles even in aggressive and transforming murine CLL. This proof-of-principle study established (TC)-NER as a mechanism to be further exploited to resensitize CLL cells. Leukemia (2017) 31, 1177–1186; doi:10.1038/leu.2016.294

INTRODUCTION Targeting of tumor-specific pathways in a more profound way than accomplished by conventional cytostatics is a current therapeutic challenge in chronic lymphocytic leukemia (CLL). Although response rates by chemoimmunotherapies based on fludarabine or bendamustine backbones are high, ensuing relapses are frequent.1,2 Despite the very encouraging results from specific small-molecule inhibitors, such as those against Bruton's tyrosine kinase, also these strategies are associated with incomplete clonal eradication and refractoriness or even aggressive transformation.3,4 Besides hypermorphic mutations in genes of those targeted pathways, major mediators of general resistance to classical apoptosis and therapy in CLL are a marked prosurvival impact by microenvironmental niches5 and genetically dictated deficiencies to evoke an adequate ATM/p53-mediated DNA damage response (DDR) or otherwise-triggered apoptotic clearance. Those are particularly found in the 11q23/ATM- or 17p/TP53deleted/mutated high-risk CLL.6 Higher eukaryotes use different mechanisms for detecting and repairing potentially harmful or protumorigenic DNA damage in actively transcribed genes and in the entire genome. The chromatin structure under low-replicative activity, such as in quiescent lymphocytes, imposes constraints on the accessibility of classical (i.e. DNA double-strand break) repair systems. Active transcription

reduces the prevalence of somatic mutations accumulating in such low-access regions, particularly by using the form of nucleotide excision repair (NER).7 NER is a complex DNA repair machinery.8 Its two distinct recognition systems detect helix-distorting DNA lesions. Global genome-NER (GG-NER) scans throughout the genome, whereas transcription-coupled-NER ((TC)-NER) specifically senses lesions on the actively transcribed strand. Upon lesion recognition, the common excision machinery removes a stretch of 30 nucleotides surrounding the damage and the gap is resynthesized.8 Importantly, TC-NER ameliorates the deleterious effects of DNA adducts that block transcription, even when they escape GG-NER. This ensures gene expression and transcript integrity.9 Inherited deficiencies in NER are linked to cancer susceptibility as well as to postnatal growth retardation and premature aging, depending on the underlying mutations.10,11 Those primarily affecting the GG-NER cause Xeroderma pigmentosum (XP), which is characterized by an ≈2000x risk increase for skin carcinomas. Defects that primarily affect TC-NER give rise to the progeric and neurodegenerative Cockayne syndrome (CS) that, in support of our rationale, shows no cancer predisposition. Higher frequencies of polymorphic NER gene variants were associated with greater risks of cancer development and more likely relapses after chemotherapy.10,11 Classical S-phase chemotherapeutics, such as crosslinking and double-strand break-inducing drugs, are of insufficient activity in

1 Laboratory of Lymphocyte Signaling and Oncoproteome, Department of Internal Medicine I, Center for Integrated Oncology (CIO) Köln-Bonn, University of Cologne, Cologne, Germany; 2Excellence Cluster for Cellular Stress Response and Aging-Associated Diseases (CECAD), University of Cologne, Cologne, Germany; 3Department of Internal Medicine III, Ulm University, Ulm, Germany; 4Institute for Genome Stability in Ageing and Disease, Medical Faculty, University of Cologne, Cologne, Germany; 5Organic Chemistry Laboratory, University of Bayreuth, Bayreuth, Germany; 6Molecular Hematology Unit, ICGEB, Trieste, Italy and 7Department of Clinical and Molecular Medicine, University La Sapienza, Roma, Italy. Correspondence: Dr M Herling, Laboratory of Lymphocyte Signaling and Oncoproteome, Department of Internal Medicine I, Center for Integrated Oncology (CIO) KölnBonn, University Hospital of Cologne, Kerpener Strasse 62, Building 15, Floor 1, Room 10, Cologne 50937, Germany. E-mail: [email protected] Received 15 May 2016; revised 1 September 2016; accepted 12 September 2016; accepted article preview online 24 October 2016; advance online publication, 2 December 2016

Targeting NER in CLL G Lohmann et al

1178

We previously described the synthesis of illudins.17,18 Trabectedin was from PharmaMar (Madrid, Spain), bendamustine was from Mundipharma (Limburg, Germany) and fludarabine from Actavis (Steinhausen, Switzerland). Detailed descriptions of mice, compound preparations, reagents, their sources, protocols for flow cytometry, PCRs (primers), immunoblots, cytomorphology, immunofluorescence, statistics and metadata analysis are given in the Supplementary Data online.

frequencies of large-fragment gains/losses were of marginal cohort percentage (0.85–3.4% in 354 CLL; ‘CLL8’ trial data) with a trend towards less frequent involvement of gene loci of the TC branch. Mutations affecting NER genes were also infrequent (gene-specific range 0.46–4.13% in 218 CLL). These data suggest that the NER pathway is functionally retained in CLL more than classical modes of DNA repair. Given the increased ROS19 and mutational burden of CLL,20,21 we propose a specific dependence of these low-replicative cells on NER. We next assessed the expression patterns of genes of the NER branches in CLL cells for their association with known risk categories as well as with response to and outcome after fludarabine-based chemotherapy. Gene expression profiling from PB-isolated CD19+ cells from 337 treatment-naïve CLL patients of the ‘CLL8’ trial of the German CLL Study Group were analyzed. This prospective phase III randomized study compared treatment with fludarabine/cyclophosphamide (FC) to FC-plus-rituximab (FCR).1,14 TC- or GG-NER- associated genes were equally expressed across subsets defined by cytogenetics, TP53 mutations or IGHV gene germline similarity (not shown). Unsupervised algorithms applied to signals from all probes for NER genes distinguished two clusters of patients (Supplementary Figure S2a). These were created independent of other established risk strata (e.g. IGHV, cytogenetics; not shown). Cluster 1 was mostly defined by upregulation of the TC-NER genes ERCC6 and ERCC8, as well as by upregulation of the common NER core factors ERCC2 and ERCC4. This subcohort of patients was associated with an inferior progressionfree survival after FC chemotherapy (Supplementary Figure S2b), a difference that was abrogated when antibody treatment was included (FCR cohort). To study differences in the expression of NER genes in fludarabinerefractory disease,22 we compared qRT-PCR-based profiles of four patients at the stage of progression under fludarabine to those with fludarabine-sensitive disease (six patients). The NER genes ERCC6L and XAB2 (all TC branch), ERCC2, ERCC4 (all common core) and RAD23A (GG) showed significantly elevated expression levels in the fludarabine-refractory cohort (Supplementary Figure S2c). Fittingly, metadata from 13 paired CLL samples at baseline vs relapse after fludarabine-based therapy revealed upregulation of ERCC8 (TC-NER), -3 and -5 (converged trunk) genes.20 Overall, these data implicate that the molecular program of CLL entails upregulation of NER factors, particularly in association with poorer (fludarabine) treatment responses. We reasoned that the (1) low-replicative activity of CLL cells, (2) the inhibitory effect of fludarabine on the NER reaction and (3) the expression of NER genes in CLL might render the leukemic cells, particularly vulnerable to NER substrate-producing agents. We further argued that a stimulated apoptotic response in CLL cells by such compounds might be especially marked when those act transcription-dependent or -blocking and when combined with fludarabine.

RESULTS High expression of NER genes in CLL and association with treatment response The expression of NER genes (Supplementary Table S2) in CLL and their involvement in copy-number alterations or mutations had not been addressed specifically. In a first meta-analysis of publically available array-based gene expression profiling data sets, we observed increased relative mRNA levels of certain NER genes in CLL cells over normal B cells (data sources and protocol in Supplementary Methods online). The patterns of altered transcript abundance in CLL were different from those of other B-cell lymphomas (Supplementary Figures S1a and b). NER genes showed a more homogeneous upregulation in CLL vs normal B cells than observed for other major repair branches (HR, NHEJ, BER and MMR; Supplementary Figures S1a and c–f). Gene-specific

Selected (TC)-NER-active agents and their specific cytotoxicity To inflict damage that requires TC-NER for removal and repair, but potentially also to induce death programs following a nonrestorative outcome, we used illudins. Their induced DNA lesions are removed in a specifically TC-NER-dependent manner.23,24 TC-NER mutants demonstrated a high sensitivity to the Omphalotus olearius mushroom sesquiterpene product illudinS and its derivative irofulven.25 We performed here an initial cytotoxicity screen in NER-deficient fibroblasts (five genotypes; Supplementary Data online) using a 12-compound illudin-derivative panel that we synthesized and described previously (compounds in Supplementary Data online).17,18 We identified here illudinM and ferrocen-illudinM (ferrocen-IM) as the most efficient and selective agents. Although also highly active here and reported by us to be specific,26 we did not pursue irofulven based on its limitations by a strong retinal toxicity.27 We then confirmed the principle of our

CLL largely because of its low-replicative activity. Therefore, we postulated that CLL might be more susceptible to transcriptionblocking genotoxic stress and consequently replicationindependent (TC)-NER appears as an ideal target for this disease. In fact, the particular activity of the nucleoside analog fludarabine in CLL is suggested to relate to its incorporation into the DNA of these low-level cycling cells via NER rather than in a replicationdependent reaction.12,13 Fludarabine terminates the polymerase reaction of strand resynthesis, hence it blocks the final gap-filling step of NER.12 Based on this inhibition of the NER reaction by fludarabine, we hypothesized that the induction of transcriptionstalling DNA lesions that require NER for removal or for lethal strand break induction might exacerbate an apoptotic response to novel NER-dependent agents or may increase the efficiency of fludarabine combinations in CLL. We present here encouraging preclinical data on the selective, efficient and p53-unrelated cell death induction in CLL by (TC)NER-dependent substances. The investigated illudins, derived from a synthesized panel, and particularly trabectedin showed a marked synergism with fludarabine. Even as single substances they performed superior in the contexts of fludarabine resistance, stromal-cell protection and in systems of abrogated ATM/P53 proficiency. Considering the positive efficacy/toxicity data from murine models of even aggressive CLL, the targeting of (TC)-NER as a novel and effective interventional strategy, particularly for high-risk and refractory CLL, warrants intensified investigations. MATERIALS AND METHODS CLL patients, cell lines and murine samples The ‘CLL8’ trial cohort and gene expression array experiments from 337 ‘CLL8’ patient samples have been described.1,14 An additional set of 65 CLL patients (details in Supplementary Table S1) provided peripheral blood (PB) samples, taken 41 month after any therapy. Sampling was performed after informed consent according to Good Clinical Practice guidelines and institutional review board-approved protocols (no. 11-319). Healthy-donor PB mononuclear cells (PBMCs) were isolated by density gradient centrifugation. Cell lines, including NER genotypes of Caenorhabditis elegans and of human fibroblasts, as well as culture methods are described in Supplementary Data online. Murine splenocytes and bone marrow (BM) were harvested, purified and cultured as described.15,16 Animal procedures were approved by local officials (no. 84-02.04.2012.A394).

Compounds

Leukemia (2017) 1177 – 1186

© 2017 Macmillan Publishers Limited, part of Springer Nature.


1179

Figure 1. (TC)-NER-active substances induce specific cell death in CLL cells independent of high-risk molecular profile or preceding therapy. CLL cells and healthy-donor PBMCs were treated in short-term suspension cultures. Cell viability was assessed via AnnexinV/7-AAD flow cytometry (compared with untreated controls). (a) IlludinM and ferrocen-IM at indicated dose ranges reduce viability (means ± s.e.m.) after 24 h rather specifically in CLL cells as compared with normal PBMCs. LD50 after 48 h did not differ significantly to 24 h (data not shown). CLL cell incubation with trabectedin revealed very low LD50 after 48 h incubation time, similar to healthy-donor PBMCs; NR, not reached. (b) CLL cells from cases categorized by karyotypes into three clinically relevant subsets were treated with 5 μM fludarabine, 20 μM bendamustine, 0.1 μM illudinM, 3 μM ferrocen-IM and 4 nM trabectedin for 48 h. Cells from patients with the high-risk deletions − 11q or − 17p showed significantly higher viability (resistance) when exposed to the conventional agents fludarabine (viabilities as means ± s.e.m.: 79.94 ± 6.75 for − 11q; 89.98 ± 4.84 for − 17p; 57.11 ± 5.93 for *others) or bendamustine (81.40 ± 4.40 for − 11q; 88.42 ± 4.46 for − 17p; 57.87 ± 4.36 for *others) as compared with those with other karyotypes (‘*others’: cases with no detectable aberrations, or isolated − 13q, or +12, or − 6q). The (TC)-NERactive compounds reduced viability in the − 11q and − 17p CLL subgroups to the same extent as in those with other aberrations: illudinM: in − 11q 41.54 ± 6.19, in − 17p 34.42 ± 4.63, in *others 36.73 ± 3.65; ferrocen-IM: 47.55 ± 10.07 vs 46.79 ± 6.20 vs 59.17 ± 5.06; trabectedin: 39.06 ± 9.44 vs 47.96 ± 10.99 vs 38.95 ± 4.56. NS, not significant. (c) CLL cells from patients who had been treated previously with fludarabineor bendamustine-containing regimens ‘(+)’ as opposed to those who did not receive any prior therapy ‘(− )’ showed an increased resistance to fludarabine and bendamustine in vitro. Viabilities (mean ± s.e.m.): fludarabine 81.33 ±4.43 in (+) vs 53.37 ± 5.84 in (− ); bendamustine 79.65 ±3.91 in (+) vs 57.43 ± 3.43 in (− ). In contrast, (TC)-NER-active compounds without reduced activity in fludarabine/bendamustinepre-treated patients: illudinM 36.99 ±3.93 in (+) vs 33.01 ± 3.60 in (− ), ferrocen-IM 50.13 ±6.53 in (+) vs 55.40 ± 4.32 in (− ), and trabectedin 42.3 ±7.25 in (+) vs 48.15 ± 4.84 in (− ). (d) (TC)-NER targeting compounds with higher activity regardless of IGHV status, while CLL with unmutated (U) IGHV genes showed increased (vs mutated (M) status) resistance to fludarabine and bendamustine in vitro. Viabilities (mean ± s.e.m.): fludarabine in (M) 51.54 ±7.81 vs 79.35 ± 4.35 in (U); bendamustine 62.26 ± 6.18 in (M) vs 78.63 ± 3.35 in (U).


Leukemia (2017) 1177 – 1186


1180 chosen candidates in systems of TC-NER-defective C. elegans as an experimental system featuring postmitotic somatic tissues28 and fibroblasts derived from CS patients who are defective in TC-NER. These were hypersensitive towards illudinM and ferrocen-IM (Supplementary Figure S3). Subsequently, we also added to our panel trabectedin (Ecteinascidin 743, Yondelis). This compound perpetrates toxic lesions through an aberrant repair reaction primarily caused by TC-NER recognition of trabectedin-induced adducts.29 There was an upregulation of (TC)-NER genes after short-term in vitro exposure to ionizing irradiation, to fludarabine or to the selected compounds in CLL-derived cell lines and primary samples (Supplementary Figure S4), which expands on published patterns of transcriptional activation.30,31 Given the mechanistic links of a NER-centric action between illudins and trabectedin, we refer to them here as '(TC)-NER-active' compounds. Pharmacotherapeutic targeting of TC-NER induces selective cell death in CLL, irrespective of high-risk molecular profiles or preceding therapies We first tested whether our (TC)-NER targeting compounds are active in CLL as single substances. While IlludinM and ferrocen-IM induced specific cell death (AnnexinV/7-AAD (7-aminoactinomycin D) flow cytometry; 10 cases) in freshly isolated CLL cells with lethal dose, 50% (LD50) at low micromolar concentrations, healthy-donor PBMCs were only mildly affected (Figure 1a). Comparable experiments with other transcription-blocking, but NER-independent compounds (α-amanitin, actinomycin-D) or with the XPB-targeting triptolide (XPB (ERCC3) is also a subunit of the transcription-initiating complex TFIIH) revealed similar effects (Supplementary Figure S5). This provides further support of the therapeutic principle of exploiting transcription in combination with NER, as addressed here, but the limited clinical potential of these agents either due to high toxicity or poor pharmacokinetics prompted us to not pursue them further. The toxic profile of trabectedin in humans is well documented,32,33 but its activity in CLL is also not described. We determined the LD50 of trabectedin in CLL suspension cultures to be ≈3–4 nM (approximately half of its LD50 in healthy-donor PBMCs). Based on these LD50s, we interrogated sensitivities to the (TC)NER-active substances across defined CLL risk subsets. We previously established that transcription-blocking lesions (the substrates of TC-NER could trigger ATM- and p53-independent DDRs.34 This indicates that TC-NER substrate lesions inflicted by (TC)-NER-targeting substances might trigger cytotoxic responses even in cell types with diminished ATM or p53 competence. We stratified 38 CLL cases according to their cytogenetic aberrations (− 11q vs − 17p vs ‘others’ (no detectable aberration, or isolated − 13q, or trisomy 12, or − 6q); Supplementary Table S1). Overall, the presence of the unfavorable genomic aberrations − 11q/ − 17p did not influence the activity of the three (TC)-NER-active substances in vitro (median viability of o50% in all three cytogenetic subsets), whereas virtually all − 17p cases were resistant (480% viability) to fludarabine or bendamustine, as were 67% of the − 11q cases (Figure 1b). Existence of

non-responders to fludarabine or to a lesser degree bendamustine in the cohort of ‘others’ (10–20%) indicated that resistance against these chemotherapeutics cannot be exclusively explained by − 17p or − 11q. Experiments in systems of defined ATM deficiency (B-lymphoblastoid cells from ataxia teleangiectasia families) corroborated the observed independence of the (TC)-NER-active compounds from conventional double-strand break repair programs. ATMnull/null lines showed greater resistance towards fludarabine, but not towards treatment with illudinM, ferrocenIM or trabectedin (Supplementary Figure S6). The high in vitro activity of the (TC)-NER-active agents was preserved in cells from patients who had been exposed clinically to fludarabine- or bendamustine-containing regimens vs treatment-naïve samples (Figure 1c). In contrast, in vitro activity of fludarabine or bendamustine was greatly diminished in the clinically pre-treated CLL. The (TC)-NER-targeting compounds also conferred high cell death inductions in three samples from CLL patients who were clinically fludarabine refractory and even became resistant to inhibitors of the Bruton's tyrosine kinase (ibrutinib and CC-292; Supplementary Figure S7). Efficacy of the (TC)-NER-active compounds was equally high in samples from IGHV gene-mutated vs -unmutated CLL, whereas fludarabine and bendamustine performed poorly in the high-risk subset of unmutated cases (Figure 1d). (TC)-NER targeting compounds show synergism with fludarabine Given the role of (TC)-NER in incorporating fludarabine during the gap-filling reaction and the encouraging results of the (TC)-NER targeting substances already at the single-compound level (above), we hypothesized functional synergisms. We exposed freshly isolated CLL cells to illudinM or ferrocen-IM, each simultaneously with fludarabine (for 48 h for sufficient intracellular rephosphorylation towards active F-ara-ATP). The combination of illudinM or ferrocen-IM at concentrations 10- and 3-fold below their LD50s, respectively, with fludarabine, led to significantly higher and synergistic (‘overadditive’) cell death induction in all CLL samples than as single substances (Figures 2a–c). These synergisms remained apparent in the subsets of − 11q and − 17p CLL. There, the addition of subefficient dosages of illudins overcame fludarabine insensitivity. Similarly, sub-LD50 dosages of trabectedin were synergistic with fludarabine, also in the − 17p subset, but not with bendamustine (Figure 2d and Supplementary Figures S8a and b). Furthermore, in splenocytes from the Eμ-TCL1 murine model of fludarabine-resistant CLL, the activity of all single substances was markedly lower, but the cooperative effect of illudins with fludarabine was observed again (Supplementary Figures S8c and d). Fitting our mechanistic hypothesis, the combinations with bendamustine only led to additive effects (not shown). Surprisingly, studies of time-order relationships revealed that fludarabine treatment before the (TC)-NER-active agents was more efficient (Supplementary Figure S9) than a ‘post’ scheduling. This suggests that an alteration in the nucleotide pools towards a substitution by the nucleoside analog fludarabine needs to occur before the induction of (TC)-NER substrate lesions.

Figure 2. (TC)-NER-active substances show synergisms with fludarabine. Simultaneous treatment of isolated CLL cells with fludarabine and with sub-LD50 concentrations of TC-NER-active agents followed by assessment of cell viability at 48 h by AnnexinV/7-AAD flow cytometry (means ± s.e.m.). Significantly reduced viabilities in (a) the non-categorized cohort when fludarabine was combined with (left) illudinM (fludarabine: 73.85 ± 3.89; illudinM 0.01 μM: 90.11 ± 2.11; combination: 47.82 ± 4.73) or with (right) ferrocen-IM (fludarabine: 73.28 ± 4.67; ferrocen-IM 1 μM: 83.47 ± 2.31; combination: 31.75 ± 3.30). CLL cells from patients (pts.) with (b) − 11q (left: fludarabine: 78.74 ± 6.78; illudinM 0.01 μM: 96.26 ± 6.96; combination: 59.23 ± 9.46; right: fludarabine: 69.88 ± 6.06; ferrocen-IM 1 μM: 84.36 ± 4.17; combination: 37.84 ± 7.64) and (c) − 17p (left: fludarabine: 89.67 ± 6.13; illudinM 0.01 μM: 91.74 ± 3.65; combination: 48.91 ± 5.35; right: fludarabine: 89.67 ± 6.13; ferrocen-IM 1 μM: 80.14 ± 6.47; combination: 28.90 ± 2.12) cytogenetic aberrations with synergistically reduced viabilities in the combinations. (d) Trabectedin: in (left) non-categorized cohort (fludarabine: 69.05 ± 5.71; trabectedin 1 nM: 81.54 ± 3.69; combination: 32.09 ± 4.33) and (right) − 17p cases (fludarabine: 93.78 ± 3.80; trabectedin 1 nM: 89.74 ± 6.62; combination: 53.73 ± 8.05) with synergistically reduced viabilities. Leukemia (2017) 1177 – 1186



1181


Leukemia (2017) 1177 – 1186


1182 The (TC)-NER-active agents induce largely p53-independent apoptosis The observations of cell death induction by the (TC)-NER targeting compounds in systems of deficient ATM/p53 suggested a desired mode of p53-independent programmed cell death. We confirmed this by biochemical readouts. In detail, immunoblots for the distal marker of apoptotic cell death, cleaved poly(ADP-ribose) polymerase (PARP), revealed prominent increases in processed PARP upon compound treatment (Figure 3a and Supplementary Figure S10). This entailed dose-dependent patterns and the

fludarabine-associated synergisms of the (TC)-NER-active agents already observed in the viability recordings (above). Strikingly and in contrast to fludarabine or bendamustine, apoptosis induced by trabectedin, illudinM and ferrocen-IM was associated with the absence of an obvious p53 activation (Figure 3a and Supplementary Figure S10). Antiapoptotic markers, for example, BCL2, remained largely unaffected. Time-lines for induction/ recruitment of the double-strand break indicator and DDR mark γH2AX (by immunoblots and fluorescence microscopy) revealed that initiated apoptosis-related DNA fragmentation was triggered

Figure 3. P53-independent apoptosis of CLL cells induced by novel (TC)-NER-active substances. (a) Immunoblots from CLL cell protein lysates (representative patient no. 21; Supplementary Table S1) after 48 h of substance exposure. Dose-dependent induction of PARP cleavage after treatment with illudinM, ferrocen-IM or trabectedin, while levels of Ser15-phospho-activated p53 remained largely unaffected. PARP cleavage induced by fludarabine and bendamustine was accompanied by induction of p53pSer15. Especially, combinations of fludarabine with illudinM or with ferrocen-IM increased PARP cleavage in a ‘more-than-additive’ manner as compared with the single substances; β-actin-normalized densitometric values (ImageJ National Institute of Health (NIH); Bethesda, MD, USA). (b) Induction of DNA-associated γH2AX foci (fluorescence microscopy) in CLL cells of patient no. 21 peaked at 12 h after trabectedin. These late foci likely indicate initial strand fragmentation as part of the apoptotic cascade or inefficient repair, as inflicted double-strand breaks are marked within minutes/ o1 h followed by markings of ongoing repair. Note that NER substrate lesions are usually not marked by γH2AX. (c) AnnexinV/7-AAD flow cytometry of substance-exposed CLL cells from representative patient no. 21 illustrates the time-related shift of the gated leukemic clone from ‘viable’ (AnnexinV − /7-AAD − at 0 h) to ‘early apoptotic’ (AnnexinV+/7-AAD − at 9–18 h) to increasingly ‘late apoptotic’ (AnnexinV+/7-AAD+ at 24 h) populations. (d) MEC1 cells treated for 48 h with indicated concentrations of illudinM (LD50: 16.2 nM; ≈6x lower than in CLL cells), ferrocen-IM (LD50: 2.81 μM) and trabectedin (LD50: 0.95 nM; ≈4x lower than in CLL cells). Viability as per AnnexinV/7-AAD flow cytometry; mean ± s.e.m. (e) Immunoblots of MEC1 lysates after 48 h substance exposure demonstrating cleavages of caspase-3 and PARP paralleling elevated γH2AX levels, most markedly in trabectedin-treated cells. (f) Freshly isolated CLL cells (six cases) cultured as suspensions (white) or with NKtert BMSC support (gray); all treated with substances at indicated concentrations for 48 h. Cell viability (means ± s.e.m.) as per AnnexinV/7-AAD flow cytometry. Only trabectedin could noticeably overcome feeder-cell-mediated protection; viability: 55.88 ± 5.78 (suspension) vs 59.07 ± 11.02 (NKtert feeder coculture). Leukemia (2017) 1177 – 1186



1183

Figure 4. Antileukemic activity of trabectedin in murine models of CLL. Activity of trabectedin at indicated concentrations and schedules in three models of syngeneically transplanted splenocytes from TCL1-initiated murine CLL. (a) Extended survival of four trabectedin-treated recipients of leukemic cells freshly explanted from a Eμ-TCL1 mouse vs three saline-injected animals (identical donor; P = 0.1). Inset: characteristic morphology in blood smear; small mature cells and smudge cells (Giemsa). (b–e) System of passaged Eμ-TCL1 line derived from founder TCL1-002.44 Trabectedin and vehicle control were administered intravenously on post-transplant days +15, +18 and +22; PB samples were taken on days +15, +16, +20, +23 and +26. (b) Treatment with trabectedin at 0.05 mg/kg (d20: P = 0.006; d23: P = 0.005; d26: P = 0.003) and at 0.1 mg/kg (d20: P = 0.0004; d23: P = 0.0008; d26: P ⩽ 0.0001) significantly (*) delayed rises of white blood cell counts (WBCs) in PB. Fludarabine (established schedule,45 intraperitoneally on days +15 through +19) without a marked effect. (c) Differences in PB lymphocyte counts between trabectedin- (0.1 mg/kg) vs control- (ctrl.) treated recipients evident at times of exponential rises in PB leukemic burden (around day +23); (means ± s.e.m.): vehicle 108.6 ± 2.06 vs trabectedin 17.18 ± 3.09. Inset: PB smear with larger cells, less dense nuclear chromatin and smudge cells. (d) Serial flow cytometry recordings on PB cells: delayed outgrowth of the CD5+/CD19+ leukemic clone in a representative trabectedin-treated animal as compared with a vehicle-treated mouse. (e) Mice treated with trabectedin at 0.1 mg/kg (n = 6) or 0.05 mg/kg (n = 6) compared with vehicle control (n = 6) showed prolonged disease-specific survival, whereas fludarabine (n = 4) did not. Logrank P-values; arrows for trabectedin schedule. (f) PB smears indicating markedly increased cellular fragility (smudge cells) after 0.15 mg/kg trabectedin in the system of p53-deleted murine CLL (scheduling in Supplementary Methods online). (g) Drop of CD19+ leukemic cells in postmortem BM (FSC/SSC lymphocyte reference gate) in the trabectedin cohort of the hTCL1A+/ − ;CD19Cre;p53fl/fl system.


Leukemia (2017) 1177 – 1186


1184 most strongly by trabectedin (Figure 3b). As characteristic for apoptosis-related γH2AX marks,35 the peak appearance of these late γH2AX foci occurred before externalization of phosphatidylserine (AnnexinV positivity; Figure 3c) and before detection of nuclear degradation (7-AAD integration). Fittingly, exposure to the (TC)-NER-active compounds resulted in a time-dependent increase in AnnexinV(+)/7-AAD( − ) (9 h), followed by AnnexinV(+)/7-AAD(+) (24 h) populations representing a typical apoptotic sequence. A notable AnnexinV( − )/7-AAD(+) fraction, indicating non-apoptotic, for example, necrotic, cell death, was not detected (Figure 3c). Isolated CLL cells from PB are mainly halted in the G1/G0 phase of the cell cycle and proliferation takes place for the most part at the niches in lymph nodes, BM or spleen. Therefore, we challenged the performance of the (TC)-NER-active compounds in MEC1 cells, which are CLL-derived, but owing to the immortalization show significant proliferation. In addition, MEC1 cells carry a mutation-inflicted truncated p53 known to mediate their fludarabine resistance,36 which was confirmed here as well (Supplementary Figure S11). In contrast, the (TC)-NER-active substances reached LD50s at low concentrations with trabectedin showing the highest activity (Figure 3d). Although fludarabine and bendamustine mediated only little induction of cleaved PARP or apoptotic effector caspase-3, trabectedin strongly did so in parallel to increased levels of γH2AX more indicative of ensuing DNA fragmentation than of ongoing inefficient repair (Figure 3e). We demonstrated earlier that direct contact between CLL cells and BM stromal cells (BMSCs) strongly increases leukemic survival.37 We therefore interrogated here the sustained protection of CLL cells by BMSC in the context of compound exposure. As indicated in Figure 3f, support by NKtert cells mitigated the drug-induced apoptosis for all substances, except for trabectedin, which was equally active in killing CLL cells irrespective of the presence of stromal cells. This pattern could not be exclusively explained by a reduced activity of the protective BMSC, because cytotoxic effects on NKtert cells were noted particularly for illudinM and ferrocen-IM, but were less pronounced for trabectedin (Supplementary Figure S12). This is important because the NKtert cells are replication-arrested with mitomycin-c before the addition of CLL for coculture, which tells us about the specificity (including potential toxicity towards resting normal cells) of the (TC)-NER-active compounds, particularly of the illudins. Taken together, this set of data indicates that the selected compounds that target and/or depend on (TC)-NER are highly active in CLL in vitro and induce a non-conventional apoptotic cell death independent of p53 activation. Among them, trabectedin showed the lowest LD50 and could overcome stroma-mediated protection of leukemic cells. Trabectedin delays leukemic progression and prolongs animal survival in various murine models of CLL In contrast to the yet ‘experimental’ nature of the illudins, trabectedin is clinically well known and carries approval status.38–40 Its particularly low LD50 in CLL in vitro, the p53independent activity and least toxicity towards BMSC, all shown here, in conjunction with its milieu-targeting properties,41 provided a rationale for its more detailed examination in vivo. We first performed a pilot experiment in a system of syngeneic transfers of splenocytes from a leukemic donor mouse of a 4F10 C57BL/6 version of the Eμ-TCL1 transgenic model of aggressive IGHV-unmutated and TP53-wild-type CLL42 (Ki67+ fractions in involved spleen: 5–10% in nodular infiltration; up to 60% in diffuse pattern). To also allow the macrophage-suppressive effect of trabectedin43 to take place, we preconditioned the recipients with a single dose (chosen according to the published data43) of ET-743 before cell transplantation. Trabectedin (over solvent control) delayed the outgrowth of the engrafted leukemic (CD5+/CD19+) Leukemia (2017) 1177 – 1186

clone (Supplementary Figure S13) and prolonged the overall survival of treated mice (Figure 4a). In a next system, we used serially in vivo passaged leukemic EμTCL1 cells from the original founder TCL1-002.44 Their ~5–10 times shorter latency (4–5 weeks) for development of a more aggressive disease in syngeneic C3H/C57BL/6 F1 hosts (splenic Ki67+ fractions 15–20%; note cytomorphologic differences in PB and spleen to never-passaged primary Eμ-TCL1 leukemic cells; Figure 4 and Supplementary Figure S14) facilitated cohort building and challenged compound performance. In fact, in this model, we commenced drug treatment upon initiation of the exponential growth phase in PB. Cohort assignments were based on equal posttransplant leukemia burden. Fludarabine, given according to protocols successfully applied to mice with primary Eμ-TCL1 tumors,45 showed hardly any effect in this system. In contrast, in both trabectedin dose cohorts the counts of PB leukocytes, their lymphocytic fraction and of the actual CD5+/19+ clone (Figure 4b–d) were markedly delayed in their rise as they were significantly lower compared with those of the control or fludarabine cohort at the given times of monitoring. Most importantly, treatment with trabectedin significantly prolonged animal survival in this model of obviously fludarabine-resistant disease. Median survival beyond treatment initiation of the control group and fludarabine-treated animals was for both 15 days, whereas median post-treatment survival of trabectedin-treated animals was 19 days for 0.05 mg/kg (P = 0.02) and 23 days for 0.1 mg/kg (P = 0.0012; Figure 4e). To more mechanistically mimic a high-risk molecular makeup, we addressed trabectedin activity in a transplant model of a TP53deleted murine CLL. We provide a detailed description of this conditional hTCL1A+/ − ;CD19Cre;p53fl/fl system elsewhere. In short, disease onset is faster and survival shorter than in the original EμTCL1 model (median survival: 31 months (N = 10) vs 53.6 months (N = 9); log-rank P = 0.0009). When never-passaged transplants were used from these mice, syngeneic hosts develop noticeable disease by rising white blood cell count after 6–8 weeks. A largecell cytomorphology, a solid-tissue-accentuated manifestation and a gradual loss of CD5 from a CD5+/CD19+ early-stage expansion indicate that programs of ensuing transformation to high-grade lymphomas are active (Figure 4f; Supplementary Figures S14a and b). Immunocompetent C57BL/6 J/N syngeneic recipients were engrafted with leukemic splenocytes from hTCL1A+/−;CD19Cre; p53fl/fl donors. The Ki67+ fraction in spleens ranged from 30 to 50%. Repeated treatment with single-dose trabectedin at stages of exponentially rising PB lymphocytes only led to a mild reduction of leukemic burden in this compartment (Supplementary Figure S15a). However, upon closer examination, PB leukocytosis in the trabectedin cohort was associated with a strikingly higher percentage of smudge cells (blood smears; Figure 4f) and with a significant reduction of CD19+ leukemic cell infiltration of BM (post-mortem flow cytometry; Figure 4g). A reduction of CD68+ spleen-resident macrophages in drug-treated cohorts was paralleled in the case of trabectedin by suppressed kinetics of CD11b+/F4/80+ circulating PB monocytes (Supplementary Figures S15b and c). Trabectedin was well tolerated by all exposed mice (no weight loss or other clinical signs). In light of its high antileukemic activity in the investigated CLL models, it is encouraging to also note that its induced myelosuppression (hemoglobin levels and thrombocyte counts) appeared moderate. Comparisons with fludarabine in this respect are not entirely valid due to the aspect of a failing marrow at terminal disease progression in these animals as compared with the reduced BM disease burden in the trabectedin cohort (Figure 4g and Supplementary Figure S16). DISCUSSION The clinical reality of prevalent resistance to conventional and to more targeted therapeutics in CLL warrants the search for © 2017 Macmillan Publishers Limited, part of Springer Nature.


1185 alternative strategies. To devise principles to bypass major modes of drug resistance in CLL (i.e. an aberrant conventional DDR) and to reprogram fludarabine refractoriness into a vulnerable state, we exploited NER as an actionable mechanism. We report on the selective and efficient execution of CLL cell death by a panel of synthesized illudin derivatives and by trabectedin, both representing substances with a (TC)-NER-centric, but differential, profile of action. Genotoxic agents induce the demise of cancer cells most effectively, when these rapidly divide. Those DNA lesions lead to replication fork stalling and DNA damage checkpoint-mediated cell death or mitotic catastrophe in checkpoint-deficient cells. As CLL typically exhibits low proliferation rates and as even postmitotic cells require TC-NER to maintain the integrity of actively transcribed genomic regions, we reasoned that targeting the NER reaction would address a specific requirement of CLL cells. Furthermore, fludarabine is likely more efficiently incorporated during (TC)-NER rather than during replication.12 Moreover, the formation of NER substrate lesions in quiescent lymphocytes by platinum compounds, otherwise of low activity in CLL, can (re) sensitize them to fludarabine.46,47 In support, reports from treated CLL patients suggest that chemotherapy exerts a stimulatory effect on the activity state of NER.48 Upregulation of NER expression might hence and in turn explain fludarabine resistance that frequently occurs in those CLL that lack high-risk genomic lesions involving ATM or TP53. Our overall rationale was also based on the observed higher than normal, poor-response predicting and fludarabine resistance-associated signatures of elevated NER gene expression in CLL. Gene collectives representing other major branches of DNA repair (HR, NEHJ, BER, MMR) were less uniformly upregulated compared with those of the NER cluster. Further corroborating were our data showing a high and selective in vitro activity of transcription-targeting compounds in CLL (Supplementary Figure S5). The apoptotic CLL cell death induced by the (TC)-NER-active substances was not paralleled by p53 activation and was unaffected in ATM/p53-deficient contexts, that is, CLL cases and genetically defined cell lines or mice. Illudins induce DNA lesions that specifically require TC-NER for recovery. We previously demonstrated that upon such transcription-blocking lesions, cells evoke DDRs that are independent of canonical DNA damage checkpoint signaling and therefore do not require ATM or p53.34 Instead a caspase-9 and BCL2-independent, JNK-ERK-mediated mechanism has been proposed.49,50 For ferrocen-IM, we cannot exclude a reactive oxygen species-mediated aspect.18,19 Trabectedin-induced lesions become cytotoxic when TC-NER is activated as this gives rise to toxic repair intermediates when the NER machinery incises the damaged strand, but is unable to remove and substitute it.29 As we also observed here, this triggers a p53-independent apoptosis.29,51,52 However, the antitumor effect of this marine product uniquely combines NER-coupled DNA damage with modulatory effects on the local microenvironment, particularly targeting the immunosuppressive and proangiogenic effects of polarized tumor-associated macrophages.41,43 Our data from the CLL/stromal-cell coculture systems as well as from the compartment shifts and macrophage kinetics in our CLL mouse models strongly implicate the relevance of this milieu-shaping effect in trabectedin’s activity against CLL. In cell line screens trabectedin showed a preference for leukemias compared with the illudin irofulven.53 We were further intrigued by the high activity of our (TC)-NERactive substances in fludarabine-refractory settings and their marked synergisms with this nucleoside analog. As the final gap-filling (TC)-NER reaction is poisoned by fludarabine, these data are consistent with the requirement of (TC)-NER for the removal of illudin-induced DNA lesions that has previously been established.23 The high single-agent efficacy of the illudins in the absence of fludarabine would argue for a non-exclusive role of TC-NER in illudin© 2017 Macmillan Publishers Limited, part of Springer Nature.

mediated cell death in CLL. The particularly overadditive effects of fludarabine with trabectedin in milieu-deprived conditions can be reconciled in a model of a complete block of the NER machinery including a stalling of TC-NER-specific damage removal by trabectedin combined with additional fludarabine-mediated inhibition of a residual (compensatory GG-NER-fed) strand resynthesis. Overall, the high efficacy of the (TC)-NER-active substances in contexts of clinical chemoresistance, ATM/p53 deficiency, stromalcell protection and fludarabine combinations makes molecules that target this unique vulnerability of CLL promising candidates for further optimization. Trabectedin may hold promise for first inman data in multidrug-resistant CLL because of its additional milieu-reprogramming impact. Such a highly desired profile becomes particularly interesting in light of indications for a strong stimulatory effect of genomic complexity through impaired damage repair towards immunogenic cell death.54 CONFLICT OF INTEREST The authors declare no conflict of interest.

ACKNOWLEDGEMENTS This work was supported by the German Research Foundation (DFG KFO-286) to MaHe (HE3552/3-2), BS, Mi Ha, HCR and CDH (HE7828/1-2). A Max-Eder Excellence Award by the German Cancer Aid (to MaHe), the CLL Global Research Foundation (to MaHe), the German Jose-Carreras foundation (to MaHe, DJCLS R 12/08) and the local CECAD initiative (to MaHe and BS) further supported this work. Trabectedin was provided by PhamaMar (Madrid, Spain); bendamustine by Mundipharma (Limburg, Germany). A Eggle (Salzburg, Austria) provided C57BL/6 Eμ-TCL1 founders. We thank N Riet for help with animal work.

AUTHOR CONTRIBUTIONS GL designed, performed, analyzed the experiments and wrote the manuscript. EV analyzed statistical data on the CLL8 trial data set and wrote the manuscript. JB performed gene expression profiling and statistical data analysis on the CLL8 trial data set. PM, NR, CP, JS, VB and CDH performed experiments and analyzed data. DGE, GK and HCR provided mice. LC provided A-T cell lines. RS and BB designed/supervised compound chemistry. MiHa and SS provided access to trial samples and clinical data. MaHe and BS conceptualized the study, designed experiments, analyzed data and wrote the manuscript.

REFERENCES 1 Hallek M, Fischer K, Fingerle-Rowson G, Fink AM, Busch R, Mayer J et al. Addition of rituximab to fludarabine and cyclophosphamide in patients with chronic lymphocytic leukaemia: a randomised, open-label, phase 3 trial. Lancet 2010; 376: 1164–1174. 2 Cramer P, Langerbeins P, Eichhorst B, Hallek M. Advances in first-line treatment of chronic lymphocytic leukemia: current recommendations on management and first-line treatment by the German CLL Study Group (GCLLSG). Eur J Haematol 2016; 96: 9–18. 3 Farooqui MZH, Valdez J, Martyr S, Aue G, Saba N, Niemann CU et al. Ibrutinib for previously untreated and relapsed or refractory chronic lymphocytic leukaemia with TP53 aberrations: a phase 2, single-arm trial. Lancet Oncol 2015; 16: 169–176. 4 Byrd JC, Furman RR, Coutre SE, Flinn IW, Burger JA, Blum KA et al. Targeting BTK with ibrutinib in relapsed chronic lymphocytic leukemia. N Engl J Med 2013; 369: 32–42. 5 Burger JA, Ghia P, Rosenwald A, Caligaris-Cappio F. The microenvironment in mature B-cell malignancies : a target for new treatment strategies. Blood 2009; 114: 3367–3375. 6 Dreger P, Schetelig J, Andersen N, Corradini P, van Gelder M, Gribben J et al. Managing high-risk CLL during transition to a new treatment era: stem cell transplantation or novel agents? Blood 2014; 124: 3841–3849. 7 Zheng CL, Wang NJ, Chung J, Moslehi H, Sanborn JZ, Hur JS et al. Transcription restores DNA repair to heterochromatin, determining regional mutation rates in cancer genomes. Cell Rep 2014; 9: 1228–1234. 8 Spivak G. Nucleotide excision repair in humans. DNA Repair (Amst) 2015; 36: 13–18.

Leukemia (2017) 1177 – 1186


1186 9 Nadkarni A, Burns JA, Gandolfi A, Chowdhury MA, Cartularo L, Berens C et al. Nucleotide excision repair and transcription-coupled DNA repair abrogate the impact of DNA damage on transcription. J Biol Chem 2015; 291: 848–861. 10 Schumacher B, Garinis GA, Hoeijmakers JHJ. Age to survive: DNA damage and aging. Trends Genet 2008; 24: 77–85. 11 Hoeijmakers JHJ. Genome maintenance mechanisms for preventing cancer. Nature 2001; 411: 366–374. 12 Rao V, Plunkett W. Activation of a p53-mediated apoptotic pathway in quiescent lymphocytes after the inhibition of DNA repair by fludarabine. Clin Cancer Res 2003; 9: 3204–3212. 13 Sandoval A, Consoli U, Plunkett W. Fludarabine-mediated inhibition of nucleotide excision repair induces apoptosis in quiescent human lymphocytes. Clin Cancer Res 1996; 2: 1731–1741. 14 Vasyutina E, Boucas JM, Bloehdorn J, Aszyk C, Crispatzu G, Stiefelhagen M et al. The regulatory interaction of EVI1 with the TCL1A oncogene impacts cell survival and clinical outcome in CLL. Leukemia 2015; 29: 2003–2014. 15 Herling M, Patel KA, Weit N, Lilienthal N, Hallek M, Keating MJ et al. High TCL1 levels are a marker of B-cell receptor pathway responsiveness and adverse outcome in chronic lymphocytic leukemia. Blood 2009; 114: 4675–4686. 16 Delia D, Mizutani S, Panigone S, Tagliabue E, Fontanella E, Asada M et al. ATM protein and p53-serine 15 phosphorylation in ataxia-telangiectasia (AT) patients and at heterozygotes. Br J Cancer 2000; 82: 1938–1945. 17 Schobert R, Seibt S, Mahal K, Ahmad A, Biersack B, Effenberger-Neidnicht K et al. Cancer selective metallocenedicarboxylates of the fungal cytotoxin illudin M. J Med Chem 2011; 54: 6177–6182. 18 Knauer S, Biersack B, Zoldakova M, Effenberger K, Milius W, Schobert R. Melanoma-specific ferrocene esters of the fungal cytotoxin illudin M. Anticancer Drugs 2009; 20: 676–681. 19 Prinz C, Vasyutina E, Lohmann G, Schrader A, Romanski S, Hirschhäuser C et al. Organometallic nucleosides induce non-classical leukemic cell death that is mitochondrial-ROS dependent and facilitated by TCL1-oncogene burden. Mol Cancer 2015; 14: 114. 20 Landau DA, Carter SL, Stojanov P, McKenna A, Stevenson K, Lawrence MS et al. Evolution and impact of subclonal mutations in chronic lymphocytic leukemia. Cell 2013; 152: 714–726. 21 Lawrence MS, Stojanov P, Polak P, Kryukov G V, Cibulskis K, Sivachenko A et al. Mutational heterogeneity in cancer and the search for new cancerassociated genes. Nature 2013; 499: 214–218. 22 Hallek M, Cheson BD, Catovsky D, Caligaris-Cappio F, Dighiero G, Döhner H et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood 2008; 111: 5446–5456. 23 NGJ Jaspers, Raams A, Kelner MJ, Ng JMY, Yamashita YM, Takeda S et al. Antitumour compounds illudin S and Irofulven induce DNA lesions ignored by global repair and exclusively processed by transcription- and replication-coupled repair pathways. DNA Repair (Amst) 2002; 1: 1027–1038. 24 Koeppel F, Poindessous V, Lazar V, Raymond E, Sarasin A, Larsen AK. Irofulven cytotoxicity depends on transcription-coupled nucleotide excision repair and is correlated with XPG expression in solid tumor cells. Clin Cancer Res 2004; 10: 5604–5613. 25 Escargueil AE, Poindessous V, Soares DG, Sarasin A, Cook PR, Larsen AK. Influence of irofulven, a transcription-coupled repair-specific antitumor agent, on RNA polymerase activity, stability and dynamics in living mammalian cells. J Cell Sci 2008; 121: 1275–1283. 26 Babu V, Hofmann K, Schumacher B A C. elegans homolog of the Cockayne syndrome complementation group A gene. DNA Repair (Amst) 2014; 24: 57–62. 27 MacDonald JR, Muscoplat CC, Dexter DL, Mangold GL, Chen SF, Kelner MJ et al. Preclinical antitumor activity of 6-hydroxymethylacylfulvene, a semisynthetic derivative of the mushroom toxin illudin S. Cancer Res 1997; 57: 279–283. 28 Mueller MM, Castells-Roca L, Babu V, Ermolaeva MA, Müller R-U, Frommolt P et al. DAF-16/FOXO and EGL-27/GATA promote developmental growth in response to persistent somatic DNA damage. Nat Cell Biol 2014; 16: 1168–1179. 29 Takebayashi Y, Pourquier P, Zimonjic DB, Nakayama K, Emmert S, Ueda T et al. Antiproliferative activity of ecteinascidin 743 is dependent upon transcriptioncoupled nucleotide-excision repair. Nat Med 2001; 7: 961–966. 30 Cramers P, Filon AR, Pines A, Kleinjans JC, Mullenders LHF, van Zeeland AA. Enhanced nucleotide excision repair in human fibroblasts pre-exposed to ionizing radiation. Photochem Photobiol 88: 147–153. 31 Rosenwald A, Chuang EY, Davis RE, Wiestner A, Alizadeh AA, Arthur DC et al. Fludarabine treatment of patients with chronic lymphocytic leukemia induces a p53-dependent gene expression response. Blood 2004; 104: 1428–1434.

32 Blay J-Y, Leahy MG, Nguyen BB, Patel SR, Hohenberger P, Santoro A et al. Randomised phase III trial of trabectedin versus doxorubicin-based chemotherapy as first-line therapy in translocation-related sarcomas. Eur J Cancer 2014; 50: 1137–1147. 33 Demetri GD, Chawla SP, von Mehren M, Ritch P, Baker LH, Blay JY et al. Efficacy and safety of trabectedin in patients with advanced or metastatic liposarcoma or leiomyosarcoma after failure of prior anthracyclines and ifosfamide: results of a randomized phase II study of two different schedules. J Clin Oncol 2009; 27: 4188–4196. 34 Garinis GA, Uittenboogaard LM, Stachelscheid H, Fousteri M, van Ijcken W, Breit TM et al. Persistent transcription-blocking DNA lesions trigger somatic growth attenuation associated with longevity. Nat Cell Biol 2009; 11: 604–615. 35 Rogakou EP, Nieves-Neira W, Boon C, Pommier Y, Bonner WM. Initiation of DNA fragmentation during apoptosis induces phosphorylation of H2AX histone at serine 139. J Biol Chem 2000; 275: 9390–9395. 36 Henrich S, Christopherson RI. Multiple forms of nuclear p53 formed in human Raji and MEC1 cells treated with fludarabine. Leukemia 2008; 22: 657–660. 37 Sivina M, Hartmann E, Vasyutina E, Boucas JM, Breuer A, Keating MJ et al. Stromal cells modulate TCL1 expression, interacting AP-1 components and TCL1-targeting micro-RNAs in chronic lymphocytic leukemia. Leukemia 2012; 26: 1812–1820. 38 Grosso F, Jones RL, Demetri GD, Judson IR, Blay J-Y, Le Cesne A et al. Efficacy of trabectedin (ecteinascidin-743) in advanced pretreated myxoid liposarcomas: a retrospective study. Lancet Oncol 2007; 8: 595–602. 39 Monk BJ, Herzog TJ, Kaye SB, Krasner CN, Vermorken JB, Muggia FM et al. Trabectedin plus pegylated liposomal Doxorubicin in recurrent ovarian cancer. J Clin Oncol 2010; 28: 3107–3114. 40 Jordan K, Jahn F, Jordan B, Kegel T, Müller-Tidow C, Rüssel J. Trabectedin: supportive care strategies and safety profile. Crit Rev Oncol Hematol 2015; 94: 279–290. 41 D’Incalci M, Badri N, Galmarini CM, Allavena P. Trabectedin a drug acting on both cancer cells and the tumour microenvironment. Br J Cancer 2014; 111: 646–650. 42 Bichi R, Shinton SA, Martin ES, Koval A, Calin GA, Cesari R et al. Human chronic lymphocytic leukemia modeled in mouse by targeted TCL1 expression. Proc Natl Acad Sci USA 2002; 99: 6955–6960. 43 Germano G, Frapolli R, Belgiovine C, Anselmo A, Pesce S, Liguori M et al. Role of macrophage targeting in the antitumor activity of trabectedin. Cancer Cell 2013; 23: 249–262. 44 Suljagic M, Longo PG, Bennardo S, Perlas E, Leone G, Laurenti L et al. The Syk inhibitor fostamatinib disodium (R788) inhibits tumor growth in the Eμ- TCL1 transgenic mouse model of CLL by blocking antigen-dependent B-cell receptor signaling. Blood 2010; 116: 4894–4905. 45 Johnson AJ, Lucas DM, Muthusamy N, Smith LL, Edwards RB, De Lay MD et al. Characterization of the TCL-1 transgenic mouse as a preclinical drug development tool for human chronic lymphocytic leukemia. Blood 2006; 108: 1334–1338. 46 Takagi K, Kawai Y, Yamauchi T, Iwasaki H, Ueda T. Synergistic effects of combination with fludarabine and carboplatin depend on fludarabine-mediated inhibition of enhanced nucleotide excision repair in leukemia. Int J Hematol 2011; 94: 378–389. 47 Zecevic A, Sampath D, Ewald B, Chen R, Wierda W, Plunkett W. Killing of chronic lymphocytic leukemia by the combination of fludarabine and oxaliplatin is dependent on the activity of XPF endonuclease. Clin Cancer Res 2011; 17: 4731–4741. 48 Barret JM, Calsou P, Laurent G, Salles B. DNA repair activity in protein extracts of fresh human malignant lymphoid cells. Mol Pharmacol 1996; 49: 766–771. 49 Herzig MCS, Trevino A V, Liang H, Salinas R, Waters SJ, MacDonald JR et al. Apoptosis induction by the dual-action DNA- and protein-reactive antitumor drug irofulven is largely Bcl-2-independent. Biochem Pharmacol 2003; 65: 503–513. 50 Wang W, Waters SJ, MacDonald JR, Roth C, Shentu S, Freeman J et al. Irofulven (6hydroxymethylacylfulvene, MGI 114)-induced apoptosis in human pancreatic cancer cells is mediated by ERK and JNK kinases. Anticancer Res 22: 559–564. 51 Moneo V, Serelde BG, Fominaya J, Leal JFM, Blanco-Aparicio C, Romero L et al. Extreme sensitivity to Yondelis (Trabectedin, ET-743) in low passaged sarcoma cell lines correlates with mutated p53. J Cell Biochem 2007; 100: 339–348. 52 Damia G, Silvestri S, Carrassa L, Filiberti L, Faircloth GT, Liberi G et al. Unique pattern of ET-743 activity in different cellular systems with defined deficiencies in DNA-repair pathways. Int J Cancer 2001; 92: 583–588. 53 Poindessous V, Koeppel F, Raymond E, Comisso M, Waters SJ, Larsen AK. Marked activity of irofulven toward human carcinoma cells : comparison with cisplatin and ecteinascidin. Clin Cancer Res 2003; 9: 2817–2825. 54 Le DT, Uram JN, Wang H, Bartlett BR, Kemberling H, Eyring AD et al. PD-1 blockade in tumors with mismatch-repair deficiency. N Engl J Med 2015; 372: 2509–2520.

Supplementary Information accompanies this paper on the Leukemia website (http://www.nature.com/leu)

Leukemia (2017) 1177 – 1186
